Service commitment
We credit the minutes you lost.
No uptime percentage appears on this page. A video conferencing platform for business either carried your meeting or it did not, and the only honest unit is minutes of unavailability measured from our own status data. Below is the whole clause ledger: what counts as an outage, how the credit is worked out, what we will not credit, and how to file.
Most vendors publish a figure with three nines in it and then write four pages of definitions that make the figure unreachable. We do the opposite. There is no percentage here because we will not promise a number we cannot measure from our own probes, and our probes run from four regions on a 60-second interval, which means the smallest honest resolution we have is one minute. Every clause in this ledger is written against that resolution.
The ledger binds whichever modules sit on your order form. If you bought Meetings and Dial-in and the recording pipeline went down, you get nothing for recording, because you do not pay for recording. That is the trade: narrow scope, real money.
Part one
What counts as an outage
| Clause | What it means in plain words | Why it applies |
|---|---|---|
| SC-01Meeting join failure | A host or a participant with a valid link cannot enter a meeting, and our probes from at least two regions fail the same join in the same minute. Each such minute counts. | Joining is the one thing nobody can work around. A degraded tile is annoying; a locked door cancels the call. |
| SC-02Media loss in an active call | Audio stops for every participant in a meeting, or video stops for every participant while audio continues, for 60 seconds or longer, traced to our media servers. | Video meeting software is allowed to drop video to protect a voice. It is not allowed to drop the voice. |
| SC-03Dial-in bridge down | PSTN numbers return busy, ring without answer or drop within 10 seconds for more than 5% of attempted calls in a 15-minute window, as logged by our carrier. | Dial-in is not a legacy feature. It is how the site foreman joins from a truck with one bar of signal, and when it fails he has no fallback at all. |
| SC-04Recording not captured | A meeting is started with recording enabled and no file is produced, or the file is unplayable. Counted per meeting, not per minute, and credited at the recording module rate. | Cloud meeting recording with transcripts is bought for the archive. A missing file is a total loss of that meeting, whatever the uptime graph says. |
| SC-05Admin console unreachable | Administrators cannot sign in to the console, change roles or export the audit log for 30 consecutive minutes while meetings continue to run normally. | Credited at a quarter rate. The platform is still carrying calls, but you have lost control of who can start them. |
| SC-06Chat and message delivery | Messages posted to a channel are not delivered to other members within 120 seconds, or file uploads under 2 GB fail, for 30 minutes or longer. | Team messaging and video calls share one licence, so a chat failure is credited against the host licence it rides on. |
| SC-07Webinar go-live failure | A scheduled webinar cannot be started by its panelists within 10 minutes of the published start time, or registered attendees are refused entry while capacity remains. | A town hall has one start time. Half an hour later is not the same event, so this clause is credited per event rather than per minute. |
| SC-08Degraded, not down | Quality drops below the tier you bought. 1080p falls back to 720p, or gallery view shrinks to the active speaker, while the meeting continues end to end. | No credit. That behaviour is the fallback working. Degrading quality to hold a call is the design, and we would rather do it than disconnect you. |
Nine claims in ten are not our outage. They are one participant's upload path. Most calls fail on the upload side. Your attendees can hear you fine; you cannot hear them. Before you file, open the status board and check whether anyone outside that one office saw the same thing. We will do that check for you anyway, and we will say so plainly when the answer is no.
Part two
How the credit is calculated
| Clause | What it means in plain words | Why it applies |
|---|---|---|
| CR-01Minutes are cumulative | Every qualifying minute in the calendar month is added together, whatever the cause and however many separate incidents it came from. | Four seven-minute failures on four Mondays hurt more than one 28-minute failure at 3 a.m. The ledger treats them the same, which is the least we can do. |
| CR-02The bands | 30 to 120 minutes credits 5% of that month's fee for the affected module. 121 to 480 minutes credits 10%. 481 minutes to 24 hours credits 25%. More than 24 hours credits 50%. | Bands, not a sliding formula, because a formula invites an argument about the third decimal place and a band does not. |
| CR-03Per module, not per account | The credit applies to the fee for the module that failed. A dial-in bridge failure credits against per-minute PSTN charges for that month; a recording failure credits against retained storage. | Price basis differs by module: per host per month, per attendee block, per minute of audio, per GB retained. One blended credit would be arbitrary. |
| CR-04Per-event clauses | SC-04 and SC-07 are counted per lost meeting or webinar. Each lost recording credits one day of that month's recording storage fee. Each failed webinar go-live credits 10% of the attendee block fee. | A minute count makes no sense for an event that never started. Counting events keeps the arithmetic checkable by you. |
| CR-05Form of the credit | Credits are applied to the next invoice in USD. They are not paid out in cash, not transferable to another customer, and not exchangeable for additional seats or minutes. | A credit against a monthly agreement you can cancel at any time is already a real cost to us. A cash refund process would be slower for you and dearer for both of us. |
| CR-06Monthly cap | Total credits in any calendar month cannot exceed 50% of that month's total fees, even where several modules failed in the same month. | Past 50% the conversation is no longer about credits. It is about whether you should still be a customer, and we will raise that ourselves. |
| CR-07Credits are the remedy | Service credits are the only remedy under this page for unavailability. They do not cover lost revenue, a cancelled launch or the cost of an audience that went home. | We price the licence, not your event. Anything wider would have to be priced into the licence, and you would pay for it every month whether or not you ever claimed. |
| CR-08Trials and pilots | No credit accrues during an unpaid pilot or an evaluation workspace. We still publish the incident and still write to you about it. | There is no fee to credit. The pilot exists so you can find out how the platform behaves on your network before money changes hands. |
A worked example
A workspace of 40 hosts on the Business tier, with recording retained for 30 days and about 9,000 dial-in minutes a month. In March the media layer in one region fails twice: 46 minutes on a Tuesday morning and 92 minutes on a Thursday afternoon. That is 138 cumulative minutes, which lands in the second band, so 10% of March's host licence fee is credited to the April invoice. The dial-in bridge stayed up throughout, so the PSTN minutes are not credited. One recording from the Thursday incident never produced a file, so SC-04 adds one day of March's storage fee on top. Two lines on the April invoice, both traceable to timestamps you can read on the status board.
What the example does not include is the hour your team spent rescheduling. We do not credit that, and we will not pretend otherwise.
Part three
What is excluded
| Clause | What it means in plain words | Why it applies |
|---|---|---|
| EX-01Your connection | Packet loss, congestion or outages on your office link, a participant's home wifi, a mobile network or a hotel. Also VPN concentrators that cannot carry media. | A 720p tile needs about 1.2 Mbps each way. Ask for 1080p and that becomes 3. Plan for the person on hotel wifi. We check the upload path with you during rollout for exactly this reason. |
| EX-02Announced maintenance | Work published on the status board at least 72 hours in advance, inside the window 01:00 to 05:00 Central Time on a Sunday, capped at four hours per calendar month. | Someone is always in a meeting somewhere. The window is narrow and capped so the exclusion cannot quietly swallow a quarter of the month. |
| EX-03Emergency security patching | Unscheduled work to close an actively exploited vulnerability, announced on the status board as it begins rather than beforehand. | Waiting 72 hours to patch a live exploit would be worse for you than the downtime. Excluded only for the duration of the patch itself. |
| EX-04Unsupported clients | Browsers more than two major versions behind current, operating systems past their vendor's end of support, and desktop apps older than 12 months. | We cannot patch a browser we do not ship. The admin console shows every client version in use so this never has to be a surprise. |
| EX-05Over the configured limit | A meeting that exceeds the participant ceiling of your tier, a webinar past its attendee block, or dial-in beyond the minute commitment on the order form. | Tell us the biggest meeting you run in a month. The licence follows from that number, not from headcount. If the real number changed, change the order form rather than hoping the ceiling bends. |
| EX-06Third-party services | Your identity provider, your calendar service, your carrier's mobile network, and any system you connect through the API or a webhook endpoint you host. | If single sign-on is down at your end, our console is still up. We will help you prove which side failed, and the audit log usually settles it in a minute. |
| EX-07Suspension for non-payment or abuse | Time during which a workspace is suspended under the terms of service for unpaid invoices or for conduct that breaches them. | Suspension is a decision, not a failure. The terms set out the notice we give before one happens. |
| EX-08Events outside our control | Regional power or network failures at a provider, government action, natural events, and large-scale internet routing incidents that take our regions off the map at once. | Named narrowly on purpose. A single datacentre failure is not this clause. We run media in four regions precisely so that one of them can fail without taking your call with it. |
Part four
How to file a claim
| Clause | What it means in plain words | Why it applies |
|---|---|---|
| CL-01Who may claim | Any administrator on the workspace, by email to [email protected] with the word credit in the subject line, or by phone on +1 (925) 555-2231 followed by an email that puts it in writing. | The audit log tells us who holds the admin role, so there is nothing to verify beyond the address the mail came from. |
| CL-02What to send | Workspace name, the date, the approximate start and end time in your local zone, the module affected, and one meeting ID if you have it. | One meeting ID lets us pull the exact media session. Without it we search by workspace and time, which takes longer and credits the same amount. |
| CL-03Window | Within 30 days of the end of the calendar month in which the outage occurred. Later than that and the probe data has rolled off our retention. | We keep status probe records for 90 days and meeting metadata for 13 months, but the minute-level detail that proves a claim is the part that expires first. |
| CL-04Our answer | A written reply within 10 working days, with the minutes we counted, the band they fall into, the credit, and the invoice it lands on. If we refuse, the reply says which clause we refused under. | A refusal without a clause number is just an opinion. You should be able to check our arithmetic against this page. |
| CL-05Credits we apply ourselves | Where an incident is wide enough that the status board shows it across a region, we calculate the credit for every affected workspace without waiting for a claim. | If we already know, making you ask is theatre. You will see the line on the invoice and an email explaining it. |
| CL-06Disagreement | Reply to the refusal and ask for the probe export. We will send the raw minute-by-minute samples for the window you named, in CSV. | The data is yours to argue with. We have changed our own answer twice after a customer read the export more carefully than we did. |
What this page deliberately does not say
It does not state an uptime percentage, because we will not publish a number we cannot measure from our own status data. It does not name a certification body or an audit we have not had. It does not promise a response time outside working hours for anything other than a total outage, and it does not promise that a credit will make a missed launch whole, because it will not.
What it does say is this: the minutes are counted from a 60-second probe in four regions, the bands are fixed, the claim window is 30 days, and the reply comes with a clause number on it. If your legal team needs the same substance in contract language, the terms of service carry it, and the admin and security page describes the audit log you would use to evidence a claim.
Size it before you sign anything
Run the configurator for your largest meeting of the month and send us the spec it writes. We answer with a written proposal: modules, licence tier, the price basis for each, and this credit ledger attached as it stands.